The preventive armamentarium against HIV/AIDS has enlarged in recent years. Changing social attitudes-an attenuation of AIDS "exceptionalism"-now permit more aggressive approaches. We propose to test the concept that network-informed community-based screening can be performed in neighborhoods at high risk for HIV transmission. We will compare three network methods-snowball sampling, chain-link sampling, and respondent driven sampling (RDS)-with regard to the "yield" of HIV- positive persons, the characteristics of these persons, and the characteristics of the networks that emerge from these sampling schemes. The underlying hypothesis is that network-informed methods will efficiently find persons at high risk because of these persons'propinquity and probability of relating to each other. In particular, these approaches will discover persons who are unaware of their infection (estimate to be about 20% of those infected) and who may therefore contribute disproportionately to transmission. Discovery of infection in a community has considerable potential for prevention. Evidence has accrued that simple awareness of HIV-positivity status may reduce high risk behavior. In addition, there is a potential to reduce transmission by more widespread use of antiretroviral therapy. Finally, such discovery permits greater application of specific (HIV counseling and behavioral change) and non-specific (drug treatment) measures to persons at high risk. Our research plan focuses on three similar but geographically distinct, communities in the city of Atlanta with a known burden of HIV. We will use one of the network approaches in each of the communities, screening with a brief questionnaire, and a rapid oral test for HIV. Those who test positive will be referred to an AIDS clinical provider for confirmation and further diagnostic and clinical workup. Persons who test negative will be requested to provide a blood specimen for antigen testing to diagnose acute HIV infection. The analysis will focus on defining the proportion of HIV-positive persons who were unaware of their infection;on the incidence and prevalence of disease in these communities;and on the differences and similarities of the networks observed by the three sampling methods. We propose to screen 1,000 persons during the two years of the project. PUBLIC HEALTH RELEVANCE: A street-based, network-driven approach to disease screen removes the screening process from traditional established sites (clinics, military installations, etc.) and places it at the point of potential transmission. If such an approach can be demonstrated to be a valid concept, and is subsequently verified by a broad-based demonstration program, it will add significantly to the armamentarium for prevention and control of HIV/AIDS.